The present invention relates to the field ofjoint transform image correlators for identifying input images of objects being viewed.
The joint transform correlator is one of the most convenient devices for correlating two images since there is no need to fabricate separate holographic filters such as the matched filter or the phase only filter. The classical joint transform correlator requires a quadratic processor in the Fourier plane. In the last decade it has been shown that it is possible to introduce nonlinearities in the image plane in order to improve the performance of these correlators. The joint transform correlator has been implemented in two ways: The first approach is based on using a camera as the square law receiver, a computer for digital image enhancement as well as for interfacing with other spatial light modulators, and finally a spatial light modulator to receive the processed data from the computer. The second approach is based on using all optical spatial light modulators.
There are two distinct types of correlators, the matched filter correlator and the binary phase only filter correlator. The matched filter correlator is less sensitive to distortions in the inputs, and therefore it performs better in the presence of noise. However, at the same time, it is insensitive to small differences between the input and target image. In contrast, the binary phase-only or BPOF correlator is highly sensitive to small changes in the input; however, this same sensitivity leads to lower performance in the presence of noise. Therefore, what is needed is a correlator that can simultaneously combine the advantages of good noise performance accompanying low discrimination, with high sensitivity associated with high discrimination.